-
ELSEVIER Materials Science and Engineering A202 (1995)
206-217
MATERIALS SCIENCE &
ENGINEERING
A
Review: aqueous tape casting of ceramic powders
D. Hotza a, P. Gre i l b
~Technische Universtitiit Harnburg-Harburg, Arbeitsbereieh
Teehnische Keramik, Denickestrasse 15, D-21071 Hamburg, Germany b
Universtitiit Erlangen-Nfirnberg, Institut fffr
Werkstoffwissenehaft en, Martensstrasse 5, D-91058 Erlangen,
Germany
Received 19 August 1994; in revised form 5 December 1994
Abstract
Slurry formulations and processing parameters of the water-based
tape casting of ceramic powders are reviewed. Additives include
binders, like cellulose ethers, vinyl or acrylic-type polymers;
plasticizers, like glycols; and dispersants, like ammonium salts of
poly(acrylic acids). Mostly alumina powders have been employed.
Hydrophobing of ceramic powders permits the aqueous processing even
of water-reactive powders, like aluminium nitride. Non-toxicity and
non-inflammability of water-based systems represent an alternative
to organic solvent-based ones. Aqueous slurries are, on the other
hand, complex multiphase systems, very sensitive to process
variations. Statistical design of experiments was used for the
improvement of the process.
Keywords: Tape casting; Ceramic powders; Slurries
1. Introduction
Tape casting is a well-established technique used for
large-scale fabrication of ceramic substrates and multi- layered
structures [1 8]. A slurry consisting of the ceramic powder in a
solvent, with addition of disper- sants, binders and plasticizers,
is cast onto a stationary or moving surface. The cast tape, with a
typical thick- ness in the range of 100-300 /~m is then dried and
finally sintered to obtain a desired final shape.
Depending on the composition of the ceramic pow- der a variety
of non-aqueous organic solvents, such as alcohols, ketones or
hydrocarbons are commonly used to prepare highly concentrated
suspensions with repro- ducible rheological properties and drying
behaviour.
In recent years, the environmental and health aspects of the
tape casting process have received special atten- tion. Therefore,
slurry formulations using water as sol- vent instead of organic
liquids have appeared in the literature [9-24]. Non-aqueous
solvents have lower boiling points and avoid hydratation of the
ceramic powder, but require special precautions concerning tox-
icity and inflammability. Typically, organic solvent re- covery
systems are needed to control emissions of compounds into the
atmosphere. On the other hand, an aqueous system has advantages of
incombustibility,
non-toxicity and low cost, associated with the large amount of
experience with the use of water in similar ceramic powder
processes, such as slip casting. In addi- tion, other colloidal
processing methods, such as those used in paint or magnetic tape
fabrication, have changed from organic to water-based systems due
to safety considerations.
A tape casting slurry must be adjusted in order to yield tapes
which satisfy some quality criteria, such as (i) no defects during
drying; (ii) cohesion to allow the manipulation of dried sheets;
(iii) microstructural ho- mogeneity; (iv) good thermocompression
(lamination) ability; (iv) easy pyrolysis (burnout); and (v) high
me- chanical strength after sintering. This requires careful
selection of the slurry additives together with accurate control of
many processing parameters.
Major differences between non-aqueous and aqueous tape casting
refer to the sensitivity to process perturba- tions, as reported by
Nahass et al. [15]. An organic solvent-based slurry is much more
volatile and irritating to process, but strong, uniform green tapes
are easy to achieve. An aqueous slurry has smaller tolerance to
minor changes in drying conditions, casting composi- tion or film
thickness. It produces crack-free, uniform green tapes only when
all variables are controlled ex- tremely well. The aim of this work
is to review the
0921-5093/95/$09.50 1995 Elsevier Science S.A. All rights
reserved SSDI 0921-5093(95)09785-6
-
D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206-217
Table 1 Aqueous slurry additives for tape casting a
207
Powder Binder Plasticizer Dispersant Others Reference
Alumina + MgO Acrylic polymer PEG + BBP Condensed aryl
Octylphenoxy- [9] sulfonic acid ethanol b
wax emulsion c
Alumina + talc PAA Glycerol NH4PMA + Dispex A40 e
PAA Glycerol + PVP NH4PM + Dispex A40 e
Alumina PUR POENPE PVA Glycerol POENPE PVAc Glycerol POENPE
NH4PA Glycerol + DBP Primal 850 e
Alumina Cellulose ether NH4PMA (MC, HPMC or HBMC)
Alumina Acrylic BBP NH 4 salt of a copolymer polyectrolyte
Alumina PVAc NaCMC
Alumina Acrylic polymer PPG NH4PMA (PEA + PMMA)
Mullite Acrylic polymer PPG (PEA + PMMA) Acrylic polymer PPG
NH4PMA (PEA + PMMA)
Alumina HEC PEG NH4PA
Alumina AE/AA AE/AA Acrylic
dispersants
Alumina PAA PEG NH4PA
Silicon organics c
Pin oil d
[lO]
[11]
[12]
[13]
[14]
[15,16]
[17]
[18]
[19,20,22,23] [21]
[241
aBBP is benzyl butyl phthalate; DBP is dibutyl phthalate; POENPE
is poly(oxyethylene nonylphenol ether); PUR is polyurethane; the
remaining abbreviations are Listed in Tables 2 and 3. bwetting
agent ~defoamer dsurfactant ecommercial name (no composition
given)
efforts conducted to develop aqueous systems as a reliable
alternative to organic solvent-based systems for the tape casting
process.
2. Slurry formulation
Compared with non-aqueous solvents, the variety of water-soluble
binders, plasticizers and dispersants is restricted to a few
systems, which will be discussed in the following sections. Table 1
summarizes the combi- nations used for aqueous tape casting of
alumina and mullite.
Some general rules can be inferred for the prepara- tion of a
tape casting slurry: (i) the ratio between organic components and
ceramic powder must be as low as possible; (ii) the amount of
solvent must be fixed
at the minimum to maintain a homogeneous slurry; (iii) the
amount of dispersant must be the minimum neces- sary to ensure the
stability of the slurry; (iv) the plasti- cizer to binder ratio
must be adjusted to make the tape flexible, resistant and easy to
release.
A compilation of compositions of aqueous slurries is illustrated
in Figs. 1 and 2. According to the first two mentioned criteria,
the top of the triangle in Fig. 1 should be the goal to be reached.
In other words, the minimal amount in water and in organic
additives to prepare a slurry with satisfactory properties should
be used. The ceramic powder charges vary from about 25 until almost
80 wt.%. The organic additives are always above 18 wt.%, while the
water content ranges from less than 20 up to 70 wt.%.
The organic components themselves have been used in quite
different ratios, as shown in Fig. 2. This
-
208 D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206-217
o 1oo
20 /~ 80
1oo. . 0 0 20 40 60 80 100
O Medowski and Sutch (72) Kemr and Mizuhara (82) DKita et al.
(82) I lGurak et al. (87) zxSchuetz et al. (87) Spauszus and Nobst
(87) ~7 Nahass et al. (90,92) Nagata (91-93) 0 Ushifusa and Cima
(91) $ Burnfield and Peterson (92) -,I(- Ryu et al. (93)
Water, wt% Fig, 1. Aqueous slurry formulations for tape casting.
The ceramic powder is alumina, with the exception of Ushifusa and
Cima [17], who used mullite.
classification, however, is not absolute: some additives have
multifunctional characteristics. A certain binder, for instance,
can exhibit plasticizing or dispersing effects. Two investigators
use neither dispersant nor plasticizer in the slurry formulations
[12,19-23]; in the other cases the binder corresponds at least to
half of the organic part of the slurry. The dispersant content is
in any case the lowest of all three organic additives ( < 20
wt.%). Plasticizers were generally used up to around 50 wt.%, but
Kemr and Mizuhara [10] have used much higher contents.
2.1. Solvent
The solvent dissolves the organic materials and dis- tributes
them uniformly throughout the slurry. It is the vehicle that
carries the ceramic particles in a dispersion until it evaporates
and leaves a dense tape on the carrier.
A non-aqueous suspension dries quickly and pro- duces green
sheets having a high density and a fine surface appearance. An
aqueous suspension has the disadvantages of high evaporation latent
heat and infe- rior drying characteristics, and there are many
quality problems to be solved.
A comparison of aqueous and non-aqueous slurries for tape
casting was made by Nahass et al. [15] to determine the effect of
changing solvent systems on slurry processing and green tape
quality. An alumina powder was utilized and the other slurry
components as well as the processing parameters were kept constant.
Green tapes from both systems had similar physical properties,
although the aqueous systems were more sensitive to process
perturbations.
2.2. Powder
A well-characterized powder is necessary to increase reliability
in ceramic processing, and in particular in aqueous tape casting.
To achieve effective particle pack- ing, the powder must have a
small particle size. How- ever, the lower the particle size, the
higher the specific surface area, which is not convenient, because
higher tape shrinkages are produced and higher concentrations of
additives are required [7].
Generally, alumina powders have been used (see Table 1),
sometimes with addition of grain growth inhibitors and/or sintering
aids, like MgO [9] or talc [10]. Aqueous tape casting of mullite
has been reported by Ushifusa and Cima [17], of yttria-stabilized
zirconia by Raeder et al. [25]. Surface area values from 2 up to 11
m 2 g 1 have been mentioned for alumina powders. Average particle
size from 0.3 up to 1.7 ~m for alu- mina, or even of 3.3 /~m for
mullite have been cited.
A limitation of using water-based systems in tape casting should
be expected to be the incompatibility with powders susceptible to
hydratation, like CaO or MgO. However, even this can be overcome
through the hydrophobing of ceramic powders [26]. By means of this
technique water-reactive powders, like A1N [27- 29], can be
processed in aqueous media.
Hydrophobing of A1N powders was performed through adsorption of
stearic acid on the particle sur- face, using cyclohexane as
solvent. Adsorption data obtained indicated a Langmuir
chemisorption isotherm. Even after 96 h leaching in water no
crystalline phase other than A1N could be detected by XRD [29].
Tapes could be cast and sintered without significant increase in
oxygen content [30].
-
D. Hotza, P. Greil / Materials' Science and Engineering A202
(1995) 206-217 209
2o 8o
100 ~ ~ ~ ~ ~ . 0
0 20 40 60 80 1 O0
O Medowski and Sutch (72) Kemr and Mizuhara (82) [] Kita et al.
(82) ,Gurak et al. (87) A Schuetz et al. (87) Spauszus and Nobst
(87) V Nahass et al. (90,92) .N.agata (91-93) 0 ushifusa and Cima
(91) $ Burnfield and Peterson (92) -~ Ryu et al. (93)
Plasticizer, wt% Fig. 2. Organic additive formulations used in
aqueous slurries for tape casting.
2.3. Binder
The binder provides strength to green tapes after evaporation of
the solvent through organic bridges between the ceramic particles.
The tapes can then be easily manipulated and retained in the
desired shapes before sintering.
Organic binders are either dissolved or dispersed in water as an
emulsion. Most soluble binders are long- chain polymer molecules.
The backbone of the molecule consists of covalently bonded atoms
such as carbon, oxygen and nitrogen. Attached to the backbone are
side groups located at frequent intervals along the length of the
molecule. The chemical nature of the side groups determines in part
which liquids will dissolve the binder. I f the side groups are
highly polar, solubility in water is promoted [31]. The polymeric
molecules of binders consist of smaller units, the monomers. The
number of monomers in a polymer is called the degree of
polymerization, DP. The number of sites on which modifications are
made in a monomer is called the degree of substitution, DS. The
molar ratio between side groups and a monomer unit is called the
average molar degree of substitution, MS.
Two groups of substances mainly have been used as binders for
aqueous tape casting of ceramics: cellulose ethers and vinyl or
acrylic-type polymers. The only remarkable exception is
polyurethane, which was re- ferred to by Kita et al. [11].
Cellulose is a natural polysaccharide formed by ring-type monomers,
which have a modified glucose structure [31]. Fig. 3 shows the
structure of a cellulose type molecule; it is represented as a
polymeric chain that is built with a number n of cellobiose units.
The latter ones consist of two anhy-
droglucose units. Each anhydroglucose ring has three free
hydroxyls that can be substituted by various side groups through
chemical reactions. The distribution of the substituent groups is
largely determined by the corresponding reaction rate of the
hydroxyls. Some- times the reactive groups are rather attached to
sec- ondary hydroxyls present in the side chains, so that it is
usual to characterize such polymers with MS instead of DS.
Cellulose ethers that have been used as additives in the tape
casting technology are listed in Table 2. They are manufactured by
the reaction at high temperatures and pressures of alkali cellulose
with, according to the desired product, methyl chloride, ethylene
oxide, propy- lene oxide, sodium monochloracetate and others. An
idealized structure for a portion of such cellulose ethers can be
obtained for a given value of DS or MS together with the general
structural formula in Fig. 3. The etherification succeeds through
substitution and/or ad- dition on a part of the cellulose
hydroxyls. A cellulose polymer with single substituent or several
ones can be formed. In the latter case, it concerns a
copolymer.
CH2-X Z
OH CH2-Y
Fig. 3. Structural formula of cellulose derivatives.
rl
-
210 D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217
Table 2 Cellulose derivatives used in aqueous tape casting a
Compound Side groups DS MS
X Y Z
Cellulose OH OH MC Methyl OCH 3 OCH 3 HEC Hydroxyethyl
OC2H4OC2H4OH OC2H4OH HPMC Hydroxypropyl methyl -OC3 H6OH OCH 3 HBMC
Hydroxybutyl methyl -OC4HsOH OCH 3 NaCMC Sodium carboxymethyl
-OCH~COONa -OCH2COONa
OH -OCH 3 OCzH4OC2H4OH OCH 3 OCH3 OH
1.5 1.5 1.5 1.5 1.0
1.5 2.5 1.5 1.5 1.0
aUsed as binders, with exception of NaCMC, used as dispersant,
X, Y, Z see Fig. 3
It is usual to divide the cellulose ethers into ionic and
non-ionic types. Ionic cellulose ethers, like NaCMC, contain
substituents with electrical charge and are rather used as
polyelectrolytes. Non-ionic cellulose ethers, like MC and HEC,
carry no charge and are mainly used as binders. Copolymers with
ionic and non-ionic substituents are ordered in the group, whose
character predominates. Because of their different solu- bility,
non-ionic cellulose ethers are further subdivided: for instance, MC
is soluble in cold water; HEC is soluble in both cold and warm
water.
The general formula for vinyl-type additives is given in Fig. 4.
The vinyls are characterized by a linear backbone consisting of
carbon-carbon bonds, with a side group (represented here by Y, X
being a hydrogen atom) attached to every other atom. When there are
two side groups (X and Y) attached to the carbon atom, they are
called acrylics. Some vinyl and acrylic- type additives used in the
aqueous tape casting process- ing are listed in Table 3. They can
also be subdivided into ionic and non-ionic polymers. The former
are the ammonium salts of poly(acrylic acids), which act as
polyelectrolytes. The latter are used instead as binders.
Binders strongly affect the rheology of the liquid phase,
increasing the viscosity and changing the charac- teristics from
Newtonian (for pure water) to pseudo- plastic in most cases. A
pseudoplastic behaviour is characterized by a decreasing viscosity
with increasing shear rate. The rheology of the solution for its
turn directly affects the behaviour of slurries formed by adding
ceramic powders and remaining organic compo- nents. The viscosity
of aqueous slurries is very much lower compared with organic
solvent slurries. Typical values are in the range of ~0.1 to ~20 Pa
s for a
--(~ CH2
n Fig, 4. General formula of vinyl and acrylic-type
derivatives.
shear rate of 50 s-~ at room temperature [22-24]. Fig. 5 shows a
viscosity-shear rate relationship for a
2% aqueous solution of HEC, a typical, strongly pseu- doplastic
binder, with different molecular weights [31]. The pseudoplasticity
of solutions is important in many technologies, including tape
casting of ceramics. A sus- pension of solid particles tends to
settle out in water if the particles are larger than 1 /~m. The
tape casting slips would not remain homogeneous if settling oc-
curred. One approach to slow down the sedimentation is to increase
the viscosity of the liquid. However, slips must be fluid enough to
be cast. To solve this problem, a pseudoplastic solution is
utilized. The sedimentation of a particle involves very small shear
rates. Under these conditions a pseudoplastic solution may have a
very high viscosity [31]. At high shear forces, as in casting a
tape, the viscosity of the slip may be several orders of magnitude
lower. Once deposited, a slip does not run and level out throughout
the tape surface.
The suitable amount of binder to be added must be determined
experimentally. When there is not enough
Table 3 Vinyl and acrylic-type polymers used in aqueous tape
casting a
Compound Side groups
X Y
Vinyl radical H - PVA Poly(vinyl alcohol) -H OH PVAc Poly(vinyl
acetate) H -OOCCH 3 PVP Poly(vinyl pyrrolidine) -H ~ NC4H 8 PAA
Poly(acrylic acid) H -COOH AE/AA Copolymer of acrylic ester b H
-COOCH 3
and acrylic acid PEA Poly(ethyl acrylate) H COOCH2CH 3 PMAA
Poly(methacrylic acid) CH 3 -COOH PMMA Poly(methyl methacrylate)
-CH3 -COOCH 3 NHaPA Ammonium polyacrylate -H COONH 4 NHnPMA
Ammonium poly(methacrylate) CH 3 -COONH 4
aUsed as binders, with exception of NH4PA, NH4PMA and NH4PMMA,
which are generally used as dispersants of PVP, used as
plasticizer. X, Y see Fig. 4 bin this example: a methyl ester, also
called methyl acrylate
-
D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217 211
1500 ~ + 52,000 g/mol
t~ ~ 4,400 g/mol i~. 1000 r \
E
N 500
o 0 2000 4000 6000 8000
Shear rate, s -1
Fig. 5. Viscosity of a 2% aqueous solution of HEC with different
molecular weights as a function of shear rate [31].
binder, the resulting green tape tends to develop cracks. When
the amount of binder is too high, on the other hand, the tapes will
contain many voids. Fig. 6 shows the dependence of green tape
strength and density on a cellulose-type binder concentration for
different alu- mina amounts [18]. Tape strength increases and green
tape density decreases with increasing binder content,
10.0
7.5 D-
5.0 ..
if) 2.5
0.0
--O--- 34 wt% AI203
~ 1203 2 4 6 8 10
5.0
o~ 9 4.0
c- o O~ 3.0
o LLI 2.0
1.0 0 2 4 6 8 10
1:3.5 Dispersant/Plasticizer
Fig. 7. Strength (top) and elongation (bottom) of alumina green
tapes with different powder charges as a function of
dispersant/plasticizer amount [18].
10.0
7.5 Q-
.c- -~ 5.0 t-
rJ~ 2.5
0.0
~O~ 34 wt% AI203
2 4 6 8
3.4
o 3.0 o)
e.- 2.6 121
2.2 2 4 6 8
Binder content, wt%
Fig. 6. Strength (top) and density (bottom) of alumina green
tapes with different powder charges as a function of HEC wt.%
[18].
showing that a compromise must always be found.
2.4. Plasticizer
Plasticizers are additives that soften the binder in the dry or
semidry state. They are organic substances with low molecular
weight in comparison with binders and are soluble in the same
liquid. After drying, binder and plasticizer are intimately mixed.
The plasticizer breaks the close alignment and bonding of the
binder molecules, thereby increasing the flexibility and work-
ability of the tape. While softening the binder, the plasticizer
tends to reduce the strength. Fig. 7 shows the results of a tensile
test for alumina green tapes with a constant binder content (7 wt.%
HEC) [18]. This mate- rial exhibits decreasing strength and
increasing elonga- tion with increased concentration of dispersant/
plasticizer mixture (ratio dispersant/plasticizer equal to
1:3.5).
Water-soluble plasticizers used in tape casting are listed in
Table 4. They generally include glycols, in a simple form like
glycerol, or as polymers like PEG or PPG. Additions of phthalates
like BBP and DBP that are common in organic solvent-based
formulations, as well as of PVP, were instead made together with
glycol- type plasticizers: PEG + BBP [9], glycerol + PVP [10],
glycerol + DBP [11].
-
212 D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206-217
Table 4 Common plasticizers used in aqueous tape casting
Compound Formula
Glycerol HOCH2CH(OH)CH2OH PEG Poly(ethylene glycol)
HO-(CH2CH20).-H PPG Poly(propylene glycol) HO (CH2CH2CH20).-H DBP
Dibutyl phthalate C16H2204 BBP Benzyl butyl phthalate C15H2oO 4
The most important effect of the plasticizer is to reduce the
gel formation temperature, Tg, at room temperature or less. This is
illustrated in Fig. 8, in which the variation of the Tg of PVA by
adding PEG is shown [32]. Tapes having gelled liquids dry much more
slowly because the liquid does not flow to the surface during
drying. Water must leave the body by diffusion within the gel
structure. However, one advantage of a gelled structure is that the
binder does not ~nigrate to the drying surface. It would if it were
carried there by the flowing liquid as it moves to the surface.
The binder plus plasticizer system cannot strongly adhere to the
casting surface after casting, and must decompose without leaving
residues. In addition, there is an optimum value of flexibility,
obtained when the correct binder/plasticizer system is selected and
the relative concentrations are properly adjusted. If the
plasticizer concentration is progressively increased to enhance the
flexibility, the porosity will decrease until the pores disappear.
A further addition results in in- creasing interparticle distances
and the green density will decrease [5,33].
2.5. Dispersant
A dispersant, sometimes also called deflocculant, wet- ting
agent or surfactant, coats the ceramics particles and keeps them in
a stable suspension in the slurry due to steric and/or
electrostatic repulsion. Both mecha-
oo
Q. E e'- ._o
Co
"0 ..
50
40
30
20
10
o
2O 40 60 80 1 O0
Plasticizer content, wt%
Fig. 8. Effect of the plasticizer content on the binder Tg [32].
Plasticizer is PEG, binder is PVA.
nisms have been widely discussed in the literature (for a
review, see Moreno [7]). A combination of both electro- static and
steric mechanisms, referred as electrosteric, was proposed to
obtain a better stabilization [34]. The electrostatic component may
originate from a net charge on the particle surface and/or charges
associated with the anchored polymer, called polyelectrolyte. In
addition, the molecular weight of polyelectrolyte and solid loading
may change the stability and rheology of aqueous slurries [35].
The most frequently used dispersants for aqueous tape casting
are polyelectrolytes. Such additives were listed in Tables 2 and 3,
respectively, as a sodium salt of a cellulose derivative, NaCMC,
and ammonium salts of poly(acrylic acids), NH4PA or NH4PMA. The use
of aryl sulphonic acid and POENPE are also mentioned as
dispersants. Further additives were reported, although their
function is not always clear, like a so-called sur- factant (pine
oil) [15,16] and a wetting agent (octylphe- noxyethanol) [9]. In
addition, defoamers were sometimes employed (silicon-based organics
[14] or wax emulsion [9]).
In aqueous systems, the variation of pH is of particu- lar
importance due to the formation of electrostatic double layers,
which can result in high surface poten- tials and repulsive Coulomb
forces. Surface charging in aqueous environment is due to
protonation or hydroxy- lation of surface hydroxide groups
resulting in positive or negative surface charge, respectively.
Measurements of zeta potential or isoelectric point (iep) of an
aqueous suspension can also give important information about its
stability regarding pH and/or dispersant amounts [17,24]. On the
other hand, mea- sured values of iep for many ceramic powders may
differ markedly due to surface impurity contents. Rao [36]
determined the iep of commercial alumina powders in dilute aqueous
dispersions in the pH range 4 to 10.
Relative viscosity [21], sedimentation velocity and volume
[17,21], or adsorption isotherms [19-23] can be determined to find
the minimum concentration of dis- persant necessary to stabilize an
aqueous suspension. As shown in Fig. 9, a minimum in the suspension
viscosity or in the sedimentation volume represents the optimum
concentration of dispersant [21]. Optionally, a maximum in the zeta
potential curve or a constant plateau in an adsorption isotherm can
be obtained.
3. Processing and equipment
3.1. Milling and mixing
Most of the reported aqueous tape casting processes are
performed through a two-stage milling/mixing pro- cedure
[9,12-14,24]. The first stage corresponds to milling, in which a
low-viscosity slurry, consisting of
-
D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217 213
200
150 u) 0 c~
> 1 O0
I~ 50
1 2 3
Dispersant content, wt%
10.0
9.5 d E
0
9.0 r- .__q
t- I11
8.5 .~
8.0
Fig. 9. Relative viscosity and sedimentation volume of an
alumina/ water suspension as a function of dispersant amount
(adapted from Ref. [211).
water, dispersant and powder (in this order) is pre- pared.
During milling agglomerates are broken and dispersants are
uniformly distributed on the surfaces of the ceramic particles. In
the second stage, mixing and homogenization occurs in which
plasticizer and binder are dissolved in the aqueous slurry.
Some modification in this standard procedure has been mentioned.
A one-step milling was carried out with all components by many
investigators [10,11,18- 23]. Other researchers have added the
plasticizer al- ready in the first stage [9,11,13]. Ushifusa and
Cima have included an ageing step (150 h) in a ball mill without
milling media as second stage [17]. The order of addition of the
components is critical, according to previous research with
non-aqueous slurries [37-39], although the active mechanisms remain
in part unex- plained. Concerning aqueous slurries, there has been
no specific study of it. Only Nahass et al. [15] and Ushifusa and
Cima [17] describe precisely the order of addition.
Milling/mixing was mainly performed in ball mills. The milling
speed was only once mentioned (60 to 120 r.p.m.) [14]. The duration
varied from a total of about 5 h to about 24 h, with diverse
distributions for the two stages [9,19]. An ultrasonic agitation
was also cited by Nahass et al. [15,16] as the first stage of
mixing, after a previous centrifuging (3500 r.p.m., > 30 min) to
re- move excess of water.
After milling/mixing, deairing can be performed by a vacuum
[11,14,24], or else by centrifugation (2000 r.p.m., 20 min) [12].
Filtering can be also applied (400 mesh) to remove large particles
and bubbles [15,16].
3.2. Casting and drying
Casting of tapes is accomplished by the relative movement
between a "doctor blade" and a support. Two solutions are possible:
either the blade moves over a fixed support or the support moves
under a fixed blade. The first technique is discontinuous and
gener-
ally used in small-scale manufacture or for laboratory
operations. The latter is a continuous one, used in the most
production-scale tape casting processes.
Most of the investigators use the continuous process for aqueous
slurries. In this case, the moving carrier is generally covered by
a polymeric film, such as poly(ethylene terephthalate) [10,23],
polyester [11,18] or polypropylene [13]. When the discontinuous
option is used, a glass plate acts as a carrier surface, on which a
releasing agent (solution of lecithin in isopropanol) can be
applied [15-17]. To provide more precise control of the slurry
thickness, a dual doctor blade system was sometimes employed
[15-17]. Casting rates ranged from 30 to 120 cm min - 1, and gate
openings up to 1.0 mm were adjusted [15,16,24].
Drying was performed with flowing air in a closed system or at
open air, from room temperature to 85 C, with relative humidity
from 50 to 70%, for 26 min to 24 h. Spauszus and Nobst observed
that the tapes tend to be fragile when the water content decreases,
and recom- mended maintaining a residual humidity of the tapes from
2 to 5 wt.% after drying [14]. Nahass et al. [16] studied the
ageing shrinkage of alumina green tapes, which was found to
correlate inversely with the amount of organic phase bound to both
organic and aqueous- based tapes. A detailed study about the drying
of water-based tapes has not yet been made, but a theoret- ical
model for the drying of tapes in general has been presented
[2].
The density and/or the porosity of green tapes can be useful in
detecting poor packing of the powder o r excessive binder content,
for example. Techniques for estimating the tape density were
presented by Mistler et al. [2], Archimedes' principle being the
most frequently employed one. Measured values of theoretical
densities for aqueous-based tapes were between 42 and 63% for
alumina [14,20-22], and from about 45 to 50% for mullite [17].
Nagata [19,20,22] has worked with pH variations in alumina/water
slurries for tape casting in the range from pH 7.5 to 10.4. The
highest value of packing density of green tapes (about 63% of the
theoretical density) was obtained at pH 7.5. Ushifusa and Cima [17]
described the appearance of mullite/water slurries and green tapes
as a function of pH, and observed flocculation below pH 7.5.
Tensile strength of alumina green tapes has also been measured
[12,18-23]. Typical values for rupture strength are 0.39 10.68 MPa
and for elongation to failure 1.2-80%.
3.3. Shaping, burnout and sintering
After drying, the tape can be released and cut for use in a
shaping procedure, like punching [15,16] or lami- nating [17]. When
the discontinuous process is used and the tapes are cast on glass
plates, a razor blade can be
-
214 D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206-217
used to remove them from the plates [15,40]. Multilay- ered
ceramic packages were produced by a thermocom- pression
(lamination) at 120 C, with a pressure of 2.5-17.5 MPa [17].
The burnout of organic slurry-based tapes has been investigated
and partially modelled [41,42]. In the case of aqueous
tape-casting, there is no rigorous study about it.
Thermogravimetric studies performed by Bur- nfield and Peterson
[18] showed that HEC binder in the presence of alumina burns out
differently from the polymer alone. In both cases, the polymer
burns out completely, but the rate and the temperature of burnout
differ.
Ryu et al. [24] performed an investigation about the rheology of
aqueous alumina slurries for tape casting concerning changes in pH
and its influence on proper- ties of green and sintered tapes. The
bulk density of the sintered bodies followed the same tendency
observed in the green bodies, Fig. 10. The variation of green and
sintered tape densities on pH is a consequence of the pH-dependence
of the rheological characteristics of the slurry, which presented
the highest viscosity at the isoelectric point, pH 2.4.
Furthermore, this isoelectric point is significantly lower in
comparison to that of the aqueous alumina suspension without
organic compo- nents, which lays at pH 7.8. This effect is similar
to that described by Cesarano and Aksay [35] for the adsorp- tion
of a polyelectrolyte dispersant on alumina.
4. Strategies for process optimization
Aqueous slurries are a complex multiphase system, very sensitive
to qualitative and quantitative changes of the components. The
production of reliable, repro- ducible aqueous-based cast tapes
requires a close con- trol of the processing parameters or rather
the identification of processing conditions to minimize vari-
ations on product quality.
/) c- a) "0
>=
n-
100
6o
8o
7o
6o
5o 0
o
- -0 - - 1600C - -e - - 1500oC
1400C
- -B - green tape
3 6 9 12
pH
Fig. 10. Relative density as a function of slurry pH for green
and sintered tapes of alumina [24].
The process optimization is dependent on many parameters, or
factors, some of which can be controlled and others that are beyond
the control of the manufac- turer. All combinatory possibilities of
varying parame- ters (full factorial design) cannot be normally
overcome due to the extensive number of necessary experiments.
Commonly, a one-factor-by-one method is used, in which one factor
is varied while all the other factors are held constant. The
drawback of this method is that the result of each experiment is
only valid at fixed experi- mental conditions, and prediction of
experimental re- sults at other conditions is uncertain. In
contrast to this method, a fractional factorial design, with
reduced number of experiments followed by a statistical analy- sis,
can be used to optimize ceramic processing, as related in recent
investigations [29,43].
In addition, a new approach can be added to the fractional
factorial design of experiments. Normally, a statistical analysis
of variance is performed on the mean values of a chosen property,
in order to identify a setting of controllable parameters that
optimize such property. It means that the focus of traditional
experi- mental design is therefore to determine and control the
sources of variation. An alternative to this picture is to
calculate a performance statistic on a property to be optimized. It
means identifying a setting of process variables that reduce the
sensitivity of the process to the sources of variation rather than
controlling them. This approach is called robust design and it has
been success- fully applied in many research areas, including
recently in ceramic processing [44].
An application to the aqueous processing for tape casting of
alumina has been developed [30]. The influ- ence of the relative
amount of slurry additives on the slurry viscosity was analysed.
The slurry components used were water, dispersant (ammonium
polyacrylate), ceramic powder (alumina), binder (hydroxyethyl
cellu- lose), and plasticizer (glycerol), added in this order. In
every run 180 g of slurry were made. The components were mixed in a
ball mill for approximately 24 h. The viscosity of the slurries was
measured at 25 C using a rotation viscometer (Rotovisco RV20,
Haake, Karls- ruhe, Germany). In addition, tapes were cast on a
glass plate, using a double doctor blade (0.70 mm for gate height
and 60 cm min- 1 for feeding rate).
The formulations employed, according to a so-called orthogonal
array L4 [45], are summarized in Table 5. The use of this
orthogonal array makes it possible to carry out four experiments
(or runs) instead of 8 for a corresponding full factorial design.
The factors A, B and C correspond to weight percentages of
dispersant, plasticizer and binder, respectively. Two levels for
each factor were chosen: a lower and a higher weight con-
centration. Four runs were performed, viscosity mea- surements for
varying shear rates were made, and the signal-to-noise ratios Z,
nominal-is-best type, as defined
-
D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217
Table 5 Experimental design for aqueous tape casting slurries
a
215
Run No. Factor levels Slurry formulation (wt.%) Viscosity (Pa
s)
A B C x M x D xp xB XL 10 S-t 20 S-1 40 S I 50 S I
Z (dB)
1 1 1 1 33.3 0.0 0.0 6.7 60.0 2.50 2.00 1.80 1.80 15.75 2 1 2 2
33.3 0.0 4.2 10.0 52.5 12.00 10.00 8.50 8.00 14.58 3 2 1 2 33.3 0.8
0.0 10.0 55.8 6.00 5.00 4.25 4.00 14.58 4 2 2 1 33.3 0.8 4.2 6.7
55.0 0.40 0.30 0.28 0.26 13.95
ax is the weight fraction respectively of ceramic powder (M),
dispersant (D), plasticizer (P), binder (B) and solvent (L). A, B
and C are the factors that correspond to x D, xp and x a. Z is the
signal-to-noise ratio
by Taguchi [46], were calculated. The function Z, whose unit is
decibels (dB), should be always maxi- mized to make the process
insensitive to variation.
The optimum viscosity can be defined as a certain value (or
range of values) to produce a stable, easy-to- flow slurry and,
consequently, a uniform, easy-to-han- dle cast tape. To find a
combination of factors that permits the achievement of this goal,
two analyses of variance were performed for the viscosity measure-
ments, respectively on average and on variation, which are shown in
Tables 6 and 7. The fundamentals of this statistical analysis can
be found in many works about robust design of experiments
[45-48].
The analysis of variance in Table 6 identifies the statistically
significant factors that affect the mean value of the slurry
viscosity. The main contribution is due to the factor C, binder
content (about 69%). The analysis of variance in Table 7 identifies
the factors that affect the variation, calculated from the
signal-to-noise ratios Z, listed in Table 5. In this case, the main
contributions to the total variation are due to the factors A and
B, dispersant and plasticizer contents (about 43% each one).
The robust design strategy is to select the proper levels of
parameters that affect variation (to reduce the process
variability) and parameters that affect the aver- age only (to
adjust the average to the target value). From Table 7 in this
example, the factor levels A1 (no dispersant) and B l (no
plasticizer) should be selected, because they yield higher average
values for Z. The decreasing values of viscosity at increasing
shear rates, characteristic of a pseudoplastic fluid, can be under-
stood as a deviation of a Newtonian fluid behaviour. A Newtonian
fluid should be expected to have the same viscosity, independent of
the shear rate. High values of signal-to-noise ratio Z are
interpreted as a low tendency to variability for viscosity
measurements. In other words, high values of Z correspond to a
slurry with a more pronounced Newtonian character, which can be
advantageous for modelling and controlling the tape casting
slurry.
Next, from Table 6, the factor C (binder content) proved to be
an excellent adjustment factor, since it
does not affect the process variability (in Table 6). It can
then be used to adjust the viscosity value. In fact, as shown in
Fig. 11, the slurry viscosity is significantly dependent on the
binder concentration. The other slurry components are responsible
for the deviations of an "ideal" curve that corresponds to the
aqueous solu- tion of the binder. For the same binder
concentration, increasing dispersant contents decrease the slurry
vis- cosity when compared with the binder solution viscos- ity.
Since a lower binder content is desirable, sufficient to
maintain a workable cast tape, the factor level C1 (6.7 wt.%
binder) should be used. In this way, the combina- tion A1, Bj, Cl
can be identified as the optimized formulation for this system.
This combination corre- sponds to experiment number 1. Frequently,
when frac- tional factorial design of experiments is used the
combination found to be the best one was not carried on in any run.
In such a case, a verification run using the optimized combination
would be necessary to confirm the statistical analysis.
In this investigation, every run produced stable slur- ries,
which could be cast to make tapes with reasonable workability
characteristics. A remarkable fact is that even in run number 1, in
which dispersant and plasti- cizer were not added to the system,
uniform slurry and tapes were produced. This has already been
observed by other investigators, as mentioned before [12,19-23].
Nevertheless, other properties of the aqueous slurry and/or of the
cast tape, like density, tensile strength or elongation, can be
used as a quality measurement to- gether with the slurry viscosity.
A common problem is that two or more optimized properties do not
always correspond to the same combination of adjustable fac- tors.
In this case, a compromise must be found to choose the best
solution.
5. Summary
The use of water-based systems represents an alterna- tive to
the widespread non-aqueous tape casting. Non- toxicity and
non-inflammability seem to be especially
-
216 D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217
Table 6 Analysis of variance on mean viscosity of aqueous tape
casting slurries a
Factor and level s m f S V F p (%)
A ~ 46.60 5.83 1 42.61 42.61 41.07 a 19.69 A 2 20.49 2.56 B~
27.35 3.42 1 9.59 9.59 9.25" 4.05 B 2 39.74 4.97 Cl 9.34 1.17 1
146.47 146.47 141.19 c 68.89 C2 57.75 7.22
Reproducibility error 12 12.45 1.04 b Pooled error 12 12.45
Total 15 211.12
as is the sum of the measured values; m is the average of the
measured values; f i s the degrees of freedom; S is the sum of
squares; V is the variance; F is the Fisher test-value; p is the
percentage contribution to variance bpooled factor into error
Significant at 95% confidence aSignificant at 99% confidence
Table 7 Analysis of variance on signal-to-noise ratios of
viscosity of aqueous tape casting slurries a
Factor and level s m f S V F P (%)
0.80 10.74 43.33
0.80 10.74 43.33
0.07 b
A 1 30.32 15.16 1 0.80 A 2 28.53 14.27 B I 30.32 15.16 1 0.80 B
2 28.53 14.27 C 1 29.70 14.85 1 0.07 C2 29.19 14.58
Pooled error 1 0.07 Total 3 1.68
as is the sum of the signal-to-noise ratios; m is the average of
the signal-to-noise; f, S, V, F and p are defined in Table 6
bpooled factor into error
10
g 1
~ 0.1
0.01 tion HEC Slurry
0.001 0.00 0.10 0.20 0.30
Binder concentration (xJx 0
Fig. 11. Viscosity of aqueous HEC solutions and slurries at
shear rate 20 s - 1 as a function of the binder concentration
[30].
advantageous . Al l p rev ious ly repor ted data demon- strate,
however , that many aspects o f the aqueous tape cast ing process
are yet to be unders tood . The use o f
add i t ives has resu l ted f rom empi r i ca l observat ions ra
ther than f rom an unders tand ing o f the phys icochem- ical p
rocesses occur r ing at the part ic le sur face and the in teract
ions between them.
Stat is t ica l ly des igned exper iments represent a power
-
ful too l in the improvement o f ceramic process ing. In aqueous
tape cast ing, in par t i cu la r , the ob ject ive o f the
research shou ld be to make the product insens i t ive to env i
ronmenta l var iab les , p roduct deter io ra t ion and manufactur
ing imper fec t ions . Th is approach can be ach ieved by us ing
robust des ign exper iments .
References
[1] J.C. Williams, Doctor-blade process, in F.F.Y. Wang (ed.),
Treatise on Materials Science and Technology, Fol. 9, Ceramic
Fabrication Processes, Academic Press, New York, 1976, pp.
173-198.
[2] R.E. Mistier, D.J. Shanefield and R.B. Runk, Tape casting of
ceramics, in G. Onoda Jr and L.L. Hench (eds.), Ceramic Processing
Before Firing, Wiley, New York, 1978, pp. 411-448.
[3] E.P. Hyatt, Making thin, flat ceramics: a review, Am. Ceram.
Soc. Bull., 65 (4) (1986) 637 638.
[4] C. Fiori and G. de Portu, Tape casting: a technique for
preparing and studying new materials, in R.W. Davidge (ed.), Novel
Ce- ramic Processes and Applications, Br. Ceram. Proc., Vol. 38,
The Institute of Ceramics, Stoke-on-Trent, UK, 1986, pp.
213-225.
[5] A. Roosen, Basic requirements for tape casting of ceramic
powders, in G.L. Messing, E.R. Fuller and H. Hausner (eds.),
Ceramic Transactions, Vol. 1, Ceramic Powder Science II, Amer- ican
Ceramic Society, Westerville OH, 1988, pp. 675-692.
[6] R.E. Mistier, Tape casting: the basis process for meeting
the needs of the electronic industry, Am. Ceram. Soc. Bull., 69 (6)
(1990) 1022 1026.
[7] R. Moreno, The role of slip additives in tape casting
technology: Part I--Solvents and dispersants, Am. Ceram. Soc.
Bull., 71 (10) (1992) 1521-1531.
[8] R. Moreno, The role of slip additives in tape casting
technology: Part II--Binders and plasticizers, Am. Ceram. Soc.
Bull., 71 (11) (1992) 1647-1657.
[9] G.O. Medowski and R.D. Sutch, cited in J.C. Williams,
Doctor- blade process, in F.F.Y. Wang (ed.), Treatise on Materials
Science and Technology, Vol. 9, Ceramic Fabrication Processes,
Academic Press, New York, 1976, pp. 173-198.
-
D. Hotza, P. Greil / Materials Science and Engineering A202
(1995) 206 217 217
[10] M. Kemr and H. Mizuhara, Flexible Ceramic Tape and Method
of Making Same, US Patent No. 4329271 (1982).
[11] K. Kita, J. Fukuda, H. Ohmura and T. Sakai, Green ceramic
tapes and method of producing them, US Patent No. 4 353 958
(1982).
[12] J.E. Schuetz, I.A. Khoury and R.A. DiChiara, Water-based
binder for tape casting, Ceram. Ind., 66 (10) (1987) 42-44.
[13] N.R. Gurak, P.L. Josty and R.J. Thompson, Properties and
uses of synthetic emulsion polymers as binders in advanced ceramics
processing, Am. Ceram. Soc. Bull., 66 (10) (1987) 1495 1497.
[14] S. Spauszus and P. Nobst, Das Foliengiessverfahren mit
w~issrig dispegierten Bindemitteln, Silikattechinik, 38 (8) (1987)
258-262.
[15] P. Nahass, W.E. Rhine, R.L. Pober, H.K. Bowen and W.L.
Robbins, A comparison of aqueous and non-aqueous slurries for tape
casting, in K.M. Nair, R. Pohanka, and R.C. Buchanan (eds.),
Ceramic Transactions, Vol. 15, Materials and Processes in
Microelectronic Systems, American Ceramic Society, Westerville OH,
1990, pp. 355-364.
[16] P. Nahass, R.L. Pober, W.E. Rhine, W.L. Robbins and H.K.
Bowen, Prediction and explanation of ageing shrinkage in tape- cast
ceramic green sheets, J. Am. Ceram. Soc., 75 (9) (1992)
2373-2378.
[17] N. Ushifusa and M. Cima, Aqueous processing of mullite-con-
taining green sheets, J. Am. Ceram. Sot., 74 (10) (1991) 2443-
2447.
[18] K.E. Burnfield and B.C. Peterson, Cellulose ethers in tape
cast- ing formulation, in M.J. Cima (ed.), Ceramic Transactions,
Vol. 26, Forming and Technology for Ceramics, American Ceramic
Society, Westerville OH, 1992, pp. 191-196.
[19] K. Nagata, Rheology and aqueous tape casting of acrylic
ester- acrylic acid copolymer dispersed alumina, in S. Hirano, G.L.
Messing and H. Hausner (eds.), Ceramic Transactions, Vol. 22,
Ceramic Powder Science IV, American Ceramic Society, Wester- ville
OH, 1991, pp. 335 340.
[20] K. Nagata, Correlation of conformation of acrylic polymers
in aqueous suspensions and properties of alumina green sheets, in
M.J. Cima (ed.), Ceramic Transactions, Vol. 26, Forming and
Technology for Ceramics, American Ceramic Society, Westerville OH,
1992, pp. 205 210.
[21] K. Nagata, Rheological behaviour of suspension and
properties of green sheet effect of compatibility between
dispersant and binder, J. Ceram. Soc. Jpn., 101 (10) (1992)
1271-1275.
[22] K. Nagata, Dynamic viscoelastic measurements of suspension
for the drying process in tape casting, J. Ceram. Soc. Jpn., 100
(11) (1992) 1352-1356.
[23] K. Nagata, Effect of functionalities of binders on
rheological behaviour of alumina suspensions and properties of
green sheets, J. Ceram. Soc. Jpn., 101 (8) (1993) 845 849.
[24] B.-H. Ryu, M. Takahashi and S. Suzuki, Rheological
character- istics of aqueous alumina slurry for tape casting, J.
Ceram. Soc. Jpn., 101 (6) (1993) 643-648.
[25] H. Raeder, C. Simon and R. Glenne, Water-based tape casting
of yttria-stabilized zirconia: influence of powder properties, pre-
sented at the conference Ceramic Powder Science and Technol- ogy,
Friedrichshafen, Germany, 11-14 September 1994.
[26] S.M. Wolfrum and J.J. Ponje~, Surface modifcation of
powders with carboxylic acids, J. Mater. Sci. Lett., 8 (1989)
667-669.
[27] M. Egashira, Y. Shimizu and S. Takatsuki, Chemical surface
treatments of aluminium nitride powder suppressing its reactiv- ity
with water, J. Mater. Sci. Lett., I0 (1991) 994 996.
[28] P. Greil, M. Kulig, D. Hotza, H. Lange and R. Tischtau,
Aluminium nitride ceramics with high thermal conductivity from
gas-phase synthesized powders, J. Eur. Ceram. Soc., 13 (3)
(1994) 229 237.
[29] D. Hotza, O. Sahling and P. Greil, Hydrophobing of
aluminium nitride powders, J. Mater. Sci., 30 (1995) 127 132.
[30] D. Hotza, R. Janssen, N. Claussen and P. Greil,
Hydrophobing of aluminum nitride powder for aqueous tape casting,
presented at the conference Ceramic Processing Science and
Technology, Friedrichshafen, Germany, 11 14 September 1994.
[31] G.Y. Onoda Jr, The rheology of organic binder solutions, in
G. Onoda Jr and L.L. Hench (eds.), Ceramic Processing Before
Firing, Wiley, New York, 1978, pp. 235-251.
[32] C.W. Nies and G.L. Messing, Effect of glass transition
tempera- ture of polyethylene glycol-plasticized poly(vinyl
alcohol) on granule compaction, J. Am. Ceram. Soc., 67 (4) (1984)
301-304.
[33] T. Ueyama and N. Kaneko, Effect of agglomerated particles
on properties of ceramic green sheets, in P. Vicenzini (ed.), High
Tech Ceramics, Elsevier, Amsterdam, 1987, pp. 1451 1458.
[34] D.J. Shaw, Colloid stability, in Colloid and Surface
Chemistry, 4th edn, Butterworth-Heinemann, Oxford, 1992, pp.
210-243.
[35] J. Cesarano III and I.A. Aksay, Processing of highly
concen- trated aqueous alumina suspensions stabilized with
polyelec- trolytes, J. Am. Ceram. Soc., 71 (12) (1988) 1062
1067.
[36] A.S. Rao, Electrokinetic behaviour of commercial alumina
pow- ders in dilute dispersions, in G.L. Messing, K.S. Mazdiyasni,
J.W. McCauley and R.A. Haber (eds.), Advances in Ceramics, Vol. 21,
Ceramic Powder Science, American Ceramic Society, Westerville OH,
1987, pp. 517-523.
[37] K.R. Mikeska and W.R. Cannon, Dispersants for tape casting
pure barium titanate, in J.A. Mangels and G.L. Messing (eds.),
Advances in Ceramics, Vol. 9, Forming of Ceramics, American Ceramic
Society, Columbus OH, 1983, pp. 164-183.
[38] L. Braun, J.R. Morris and W.R. Cannon, Viscosity of tape
casting slips, Am. Ceram. Soc. Bull., 64 (5) (1985) 727-729.
[39] W.R. Cannon, R. Becker and K.R. Mikeska, Interactions among
additives used for tape casting, in M.F. Yan, K. Niwa, H.M. O'Bryan
and W.S. Young (eds.), Advances in Ceramics, Vol. 26, Ceramic
Substrates and Packages Jor Electronic Applica- tions, American
Ceramic Society, Westerville OH, 1989, pp. 525-541.
[40] E.P. Hyatt, Continuous tape casting for small volumes, Am.
Ceram. Soe. Bull., 68 (4) (1989) 869 870.
[41] M.J. Cima, J.A. Lewis and A.D. Devoe, Binder distribution
in ceramic greenware during thermolysis, J. Am. Ceram. Soc., 72 (7)
(1989) 1192 1198.
[42] H.-W. Yan, W.R. Cannon and D.J. Shanefield, Evolution of
carbon during burnout and sintering of tape-cast aluminium nitride,
J. Am. Ceram. Soe., 76(1) (1993) 166 172.
[43] M.W. Weiser, D.N. Lauben, P.G. Madrid and K.B. Fong,
Ceramic processing using designed partial factorial experiments,
Ceram. Eng. Sci. Proc., 14 ([ 2) (1993) 344 359.
[44] M.W. Weiser and K.B. Fong, Applications of the Taguchi
method of experimental design to improving ceramic processing, Am.
Ceram. Soc. Bull., 73 (1) (1994) 83-86.
[45] J. Krottmaier, Versuchsplannung--Der Weg zur Qualit~it des
Jahres 2000, 2nd edn, Verlag Industrielle Organisation/Verlag TIJV
Rheinland, Zurich/Cologne, 1991.
[46] G. Taguchi, System of Experimental Design, Vols. 1 and 2.
UNIPUB-Kraus/ASI, New York/Dearborn MI, 1988.
[47] M.S. Phadke, Quality Engineering Using Robust Design,
Prentice- Hall, Englewood Cliffs NJ, 1989.
[48] P.J. Ross, Taguchi Techniques for Quality Engineering,
McGraw- Hill, New York, 1988.